/^/ ***>- / s <s
\.366 Fi!e No - 1-0001
CIVIL AERONAUTIC^ BOARD
m: January 11, 1963 RELEASED? January 15, 1963
AMERICAN AIRLINES, INC., BOEING 707-123B
N 7506A, JAMAICA BAY, LONG ISLAND, NEW YORK,
MARCH 1, 1962
On March 1, 1962, American Airlines Flight One, a Boeing 707-123B, U. S.
Registry N 7506A, crashed into Jamaica Bay slightly less than two minutes
after takeoff from New York International Airport, Jamaica, New York. The
aircraft was totally destroyed. All occupants, 87 passengers and the crew
of 8, sustained fatal injuries.
Flight One was cleared for takeoff from Runway 31L on a regularly scheduled
nonstop flight to Los Angeles, California, and became airborne at 1007 e.s.t.
The bakeoff and initial climb appeared to be normal and a gentle turn to the
left was started about 3,000 feet down the runway near taxiway AA, at an altitude
of about 100 feet* Straightening out from this turn the aircraft continued to
climb for several seconds on a magnetic heading of 290 degrees, and started a
second turn to the left, apparently in compliance with radar vector directions
given by Departure Control. In the second turn the airplane continued to climb.
After initiation of the second turn the angle of bank increased until the airplane
rolled through 90 degrees of bank at a peak altitude of about 1,600 feet m. s.l.
It then entered an inverted, nose -low attitude and plunged earthward in a nearly
vertical dive. The airplane struck the shallow waters of Pumpkin Patch Channel
of Jamaica Bay approximately three miles southwest of the Idlewild Control Tower
at 1008:49 Floating debris and fuel ignited a few minutes later and burned
The Board determines that the probable cause of this accident was a rudder
control system malfunction producing yaw, sideslip and roll leading to a loss
of control from which recovery action was not effective.
On February 28, 1962, the aircraft, a Boeing 707-123B, U. S. Registry
N 7506A, was flown from Tulsa, Oklahoma, to New York International Airport,
Jamaica, New York, as American Airlines Flight 2098, a second 'section of the
regularly scheduled Flight 98. The airplane arrived at Idlewild at 0007i/
on March 1, 1962, The crew of this flight later testified that the aircraft
performed normally throughout the flight.
/ 1/ All times herein are Eastern standard time based on the 24-hour clock
- 2 -
After arrival at Idlewild and in preparation for its scheduled departure
as Flight One at 094-5 on the same day, a layover check and an origination check
were accomplished on the airplane. To correct pilot -reported discrepancies, $
a VHF receiver and the cabin pressure auto controller were changed. Cracks 4
were found in the inlet guide vanes of engine Nos. 1 and 2 and were voided. A
scheduled main oil screen change on engine No. 3 was performed,, the flight »
engineer's instrument panel light rheostat -transformer was replaced, and the
airplane was serviced. Investigation of the maintenance and servicing performed ',
on N 7506A during its layover at Idlewild showed that these tasks had been
properly completed and signed off in accordance with American Airlines procedures *
before N 7506A was released for dispatch. \
The crew involved in this accident departed Los Angeles, February 27, as
the crew of American Flight 36, flying a Boeing 720B, arriving in Boston at
0608 February 28. After a layover of about 25 hours the flight crew, operating
a Boeing 720B aircraft, American Flight 117, departed Boston on March 1 at 0720
and arrived at Idlewild at 0813. The airline dispatcher, Who- assisted the crew *
in pref light preparations for Flight One, testified that they ,T appeared to be
rational and normal." All crew members were currently certificated in compliance*
with FA A. regulations and qualified m accordance with the carrier policy and
procedure. A ground crewman involved in preparing the aircraft for departure
testified that each member of the flight crew occupied his normal crew position
for the flight. The crew consisted of Captain James T. Heist, First Officer
Michael Barna, Jr., Second Officer Robert J. Pecor, Flight Engineer Robert J. Cam;
and Stewardesses Shirley Grabow, Lois Kelly, Betty Moore, and Rosalind Stewart; .
A study of the dispatching procedures utilized for Flight One revealed that
dispatching was normal and in accordance witn stardard company procedures. The
aircraft was released from Idlewild with a total takeoff gross weight of 24,7*038 u
pounds and the center of gravity at 24-4- percent MAC (Mean Aerodynamic Chord),
both within prescribed limits. According to takeoff and climb computations the
following performance factors were to be used during the flight:
Stabilizer trim - I-l/2 units nose-up
Takeoff ground roll - 4*4-00 feet
Time to 100 knots - 23 seconds
Vl 2/ - 136 knots
V-, 2/ - 144- knots
V2 4/ - 157 knots
2/ Vj - The speed below which an outboard engine failure dictates discon-
tinuance of the takeoff and at, or above, which the takeoff may be safely cont-inuec
3/ Y-o - The lowest speed at which the pilot should apply force on the control
column to rotate the airplane for lift-off.
U v 2 " Takeoff Safety Speed; The minimum speed permissible at a height of I
35 feet, assuming engine failure at V x and rotation at Vr speed.
- 3 -
Pertinent normal operating procedures after takeoff and gear up prescribed
by the then current AAL 707-123B Operating Manual are given below for reference
purposes in reviewing this reports
Establish positive climb in straightaway flight, and accelerate to
V 2 / 20K
Order "Flaps 20" (if 30 used for takeoff)
If maneuvering is required during climb, increase speed to V"2 / 30K
before initiating turns.
After reaching 2,000 feet AFL: Accelerate to V"2 / 50K, then order
Complete terminal area maneuvering at V 2 / 50K with Flaps up. As soon
as conditions permit, accelerate to normal climb speed (300K IAS).
Section 3, page 34- stated: "Close adherence to this procedure will
result in attaining the desirable 2,000-foot altitude over critical noise
areas as quickly as possible while maintaining approximately a 30 percent
margin above stall speed fat least 4.0 knots) even though maneuvers in-
volving banks as high as 25-30° may be required while clearing terminal
Item 4. of the After Takeoff Cockpit Ch&cklist required the yaw damper to
be turned ON.
In the Boeing 707-123B aircraft the outboard ailerons are operative during
flaps down flight only. A flap system operated mechanism gradually locks out,
(fixes at neutral) the outboard ailerons when the wing flaps are retracted from
20 bo zero degrees 9 and gradually brings them into play as the flaps are extended
to 20 degrees. The purpose of this is to provide increased lateral control
during low speed operation and decreased wing torsion at high speeds when lateral
control, produced by spoilers and inboard ailerons, is ample with the outboard
ailerons deactivated. It is pertinent also to note that activation of the speed
brakes lever to the 20-degree position increases the maximum possible spoiler
displacement on the desired wing from 4-0 to 60 degrees, providing additional
A typical departure flightpath chart (Attachment 1) was prepared based on
the carrier's normal operating procedures and the computed performance capabilities
of the 707-123B. Based on a correlation of witness statements, the control tower
transcript and flight recorder data (Attachment 2) , the following sequence of
events 'occurred during the flight of N 7506A.
At 0954 Flight One, with 95 persons abroad was given taxi instructions to
Runway 31L. The flight was issued an IFR clearance nonstop to Los Angeles
International Airport at 1002. The clearance contained local departure procedures
and included the then prescribed statement ". • . in the interest of noise
abatement do not delay turn to heading two niner zero. 11 In the runup area
adjacent to Runway 31L, at 1005:05? American One advised Idlewild Tower, "ready
for takeoff," and was immediately cleared for takeoff. Flight One then taxied
5/ The Boeing Operations Manual permitted flap retraction from 30 to 20
degrees at V"2 / 10K.
- 4 -
to trie runway and was aligned with it at 1006:29. At 1006:51 the lover requested
Flight One to "advise rolling" and one second later, while ?n the taxeoff roll,
the flight so advised the tower. What appeared to De a normal takeoff was
executed and N 7506A was seen by one of the tower controllers &o lift off in
tne vicinity of taxivay "M'S approximately 5,000 feeo down the runway, at 1007,
as recorded by the controller.
At 1007: y 3 the aircraft started a gentle turn to the left in the vicinity
of taxiway "AA" (approximately 8,000 feet down the rur\;ay) at an altitude of
about 100 feet, and was established on a heading of 290 degrees at 1007; 42.
The departure controller made radar contact witn the aircraft and observed it
roll out on the 290-degree heading. At 1007*^8 the local controller advised
American One to contact Departure Control on 323.9 mcs. This call was
At 1007:54- the airplane startad a second turn to the left, and at 1008:01
American One transmitted his call sign 5 indicating that he was standing by on
Departure Control frequency. At 1003:02 the controller advised Flight One
to continue a left turn tc one four zero degrees and to report out of 2,000.
This advisory was acknowledged at 1008:09. According to the controller the
flight continued in -sfhat appeared up be a normal left turn, as seen on the
radar scope, which gives no indication of aircraft attitude. Tne target was
last oeserved ir the vicinity of the crash site. The controller continued to
issue normal radar vec turing advisories, but received no further replies from
American One* and the bar get old not reappear.
Company personnel familiar with the voices of the flight crew, after
listening to the control tower recording of transmissions from Flight One,
believed thai taey were made uv the second officer. No indications of alarm or
any abnormality on the part zf the crew were discernible during any of Flight
One ' s t ran smi j s x ons .
At 1008:23 &i unniod listed signal of one-half second duration was received
on the Departure Control frequency the sound of this signal was very sunilar
to the uf- modulated carrier associates with previous transmissions from Flight
Toe air^rafu struck ohe sa^tn in the shallow waters of Pumpkin Pabeh
Channel of Jamaica Bay during low tide wneu the depth oP cue water varied from
several inches to several feet and higher elevations of the bottom were exposed.
The geographical position of impact was fixed at 4-0°37«l ! Torth Latitude and
73 50.1" tfest Longitude, approximately three nautical miles southwest cf the
Idlevild Cont-ol Tower, Impact was made at an angle ci" approximately 78
degrees nose down, on. a magnetic heading of 30C degrees. Readings of 1 re
Fordnam University S^ismograpmc Station established tne impact time a» 1008:49.
Floating debris and fuel ignited a few minutes later aru burned fiercely. All
persons on board were fatdlly injured.
The weather at the tune of takeoff was: 15 9 000 feet scattered? visibility
15 miles | wind northwest at 1° knots,- temperature 30 degrees F$ dewpomt 11
degrees F; altimeter 30.30 menes Hg. Although the carrier' s meteorologist
testified that from the air soundings and known winds there could have been
occasional moderate turbulence ? testimony of pilots who had flown in the area
at the approximate tune of the accident indicated that they had experienced
negligible amounts. However, the vertical acceleration trace made by the
flight recorder on N 7506A indicated that Flight One did encounter light
friction turbulence. This type of turbulence was also indicated on the flight
recorder traces of other aircraft departing Idlewild at the approximate time
of Flight One's departure.
Testimony of eyewitnesses indicated that the takeoff and climb appeared
to be normal until the bank to the left steepened to an angle beyond that
usually expected of a departing aircraft. As it continued its flight, however,
witnesses observed N 7506A. continue its roll to the left through a 90-degree
bank, enter an inverted nose-low attitude and plunge downward in a dive which
was almost vertical. During the vertical portion of the dive, which was
estimated by a witness to start at approximately 900 feet, little, if any,
rotation of the wings about the longitudinal axis of the airplane was observed.
Two witnesses described a slight, abrupt bank to the left, followed by a
momentary levelling of the wings, immediately preceding the airplane's final
3teep roll and nose-over. Several witnesses testified they observed cessation
of smoke trails from two or more engines when the aircraft was nearing its peak
altitude. Only one of the witnesses registered any impression of smoke and/or
flames from unusual points of origin during the flight; this witness observed
the airplane only one or two seconds before impact. None of the witnesses
believed they saw or heard an explosion prior to impact. The majority did not
hear any unusual engine sounds, and none saw any object separate from the
aircraft during its flight. Those who paid particular attention to the an plane's
configuration were of the opinion that landing gear and wing flaps were
retracted. Except for one witness who testified that the aircraft appeared
to stall just before nosing over, most of the witnesses stated that the entire
maneuver was characterized by a smooth, continuous movement, with no indication
of recovery action being discernible.
Runway 31L is 14-, 600 feet long and 150 feet wide, with a gradient of
minus *01 percent. It was dry at the time of takeoff. The field elevation
is 12 feet m. s.l. The northwest shore line of Jamaica Bay is about 200 yards
to the left of and parallel with the runway. Heavily populated areas lie
directly beyond the end of Runway 31L.
Noise abatement procedures in effect at the time of the accident required
that flights from bhis runway not delay in making a left turn to a heading of
290 degrees after takeoff and continue the climb to 800 feet on. this heading.
After reaching 800 feet of altitude, in the interest of traffic separation,
flights were turned further left to a heading of 160 degrees and the climb was
continued on this heading for two minutes. This departure procedure therefore
accomplished two purposes: avoidance of densely populated areas, and separation
of Idlewild Runway 3XL departures from inbound traffic to LaGuardia. The
20-degree turn from Runway 31L to 290 degrees is within the limits of the jet
transport performance criteria developed for noise abatement maneuvers by
representatives of the Air Transport Association, the Air Line Pilots' Associa-
tion, and Aerospace Industries Association, and promulgated by the Federal
Aviation Agency in their Manual of Noise Abatement Procedures. Since the
accident the FAA restricted the commencement of the first turn until the aircraft
reaches an altitude of 300 feet and also eliminated the advisory, "In the
interest of noise abatement, do not delay turn to 290 degrees" from the departure
- 6 -
clearance for Runway 31L. As of December 25, 1962 the procedure was changed to
require a climb on a 290-degree heading to 1,000 feet before further turns
Insofar as possible during the investigation attention was directed to
pathological, histological, and toxicological studies of the bodies of the
flight crew which might reveal some indication of physical incapacitation.
The results of the toxicological studies were conclusive in ruling out the
possibility of incapacitation due to toxic gases, alcohol, and drugs. However,
the massive destruction of the bodies and the lack of vital portions of tissue
made it impossible to obtain results which would give irrefutable positive or
negative proof of incapacitation insofar as the pathological and histological
examinations were concerned. Studies were also wade of the medical histories
of the flight crew, but no evidence could be found to indicate that any member
had physical characteristics likely to result in an incapacitation of any kind.
£xamina.tion of the bodies of the passengers did not yield information of any
significance as to the cause of the accident except to support other evidence
that there had been no fire or explosion m flight.
Despite extensive damage to the flight recorder foil, exhaustive efforts
resulted in restoring it to a condition whereby an accurate readout could be
obtained, with two exceptions. Both exceptions involve the heading trace
as it passed through an area of tears, wrinkles and abrasions; where the
heading changed from 24,5 to 225 M, and again from 214 degrees through an
indicated reversal at 332 degrees. Only the first exception is of significance
to this investigation and will be discussed later in the report.
A chart of the flight recorder data, with added information, is attached
for reference (Attachment 2) . In discussing the recorder readout it must be
recognized that absolutely precise time correlation of the four traces was
impossible and errors of one or two seconds between them may exist. An
additional error is introduced by the friction and play in the recorder,
resulting in short-period (up to two seconds) aberrations of airspeed, altitude
and heading traces from the smoothly varying changes made by the airplane.
Characteristic of these are the numerous steps in the airspeed trace for the
takeoff run and acceleration to 190 knots. Although times, speeds, altitudes
and headings are usually expressed in terms of exact values throughout the
flight recorder discussion, it must be understood that they are approximate
but generally accurate to within plus or minus one second. The time scale
used in Attachment 2 is based on the time of impact as established by the
Fordham University seismograph. The computed aircraft performance factor,
"Maximum time to 100K," indicates the maximum number of seconds allowed for
the aircraft to accelerate to 100 knots IAS and is a measure used by crews as
acceptable performance to this point.
Except where otherwise specified, the stall speeds referred to herein
are based on a takeoff gross weight of 247,000 pounds and a gross weight at
impact of approximately 24-6,000 pounds, with a CG position of approximately
24 percent, and were determined in accordance wroh applicable type certification
requirements of Civil Air Regulations Part 4b and Special Civil Air Regulation
Study of the four traces indicates that the flight was normal during the
takeoff and the first part of the climb. The acceleration trace variation
- 7 -
from lift-off to 1008:29 indicates that N 7506A was operating in lightly choppy
air, characteristic of mechanical friction turbulence.
During takeoff , dips in the airspeed and altitude traces indicate that
rotation occurred at 1007:24 when the indicated airspeed was 145 knots
(V^ = 144 K computed). Lift-off occurred at 1007:28 when the indicated airspeed
was 155 knots (v"2 = 157 K computed). The variation in heading trace for six
seconds after lift-off corresponds with normal correction for wind drift.
The heading trace indicates that the airplane started its first turn to the
left after takeoff at 1007:37 at an indicated altitude of about 80 feet and an
airspeed of 180 knots. At this indicated altitude the airplane would still be
in its ground effect and the actual altitude would be 100 to 120 feet.
The turn continued to a heading of 290 degrees, an altitude of 260 feet
and an airspeed of 190 knots at time 1007:42. The rate of turn to this heading
corresponds to a bank angle of about 30 degrees in a coordinated turn. However,
the median acceleration trace corresponds to a bank of 20 to 25 degrees, in-
dicating a slight skid. With flaps at 20 degrees, the airplane would then be
more than 40 knots above stall. A 290-degree heading was maintained for 12
seconds (1007:42 to 1007:54) to an altitude of 700 feet with airspeed still at
At 1007:54 a second climbing turn to the left was started. This turn
continued at a nearly constant rate of 2.34 degrees/second, to a heading of
275 degrees, an altitude of 920 feet and an airspeed of 192 knots at time
1008:01. In a coordinated turn, this heading change rate corresponds to a
bank angle cf 22 degrees with the airspeed 50 to 60 knots above stall. Almost
coincidental with the end of this segment of the turn, Departure Control advised
Flight One to continue ils turn to 140 degrees.
During the same segment of the second left turn it is most likely that
the wing flap retraction from 20 to zero degrees was initiated at about 1007:57,
at an altitude of 800 feet and airspeed of 190 knots. Retraction normally
requires 12 seconds and this operation would therefore have ended at 1008:09
at an altitude of 1,350 feet and an airspeed of 200 knots. When flaps are
retracted from 20 to zero degrees it is common for the recorded airspeed to
increase slightly. During this period the trace indicates such an increase
from 190 knots to 200 knots. In addition, Boeing flight tests to simulate
Flight One as closely as possible indicate that later retraction of flaps to
zero degrees would result in lower climb out performance than reflected in
Attachment 2. Coincidental with completion of flap retraction, Flight One's
last recorded message was received by Departure Control.
Returning to consideration of the second left turn, the heading change
becomes slightly wavering from 1008:01 to 1008:07, and in the turn the heading
changed from 273 to 250 degrees, averaging a rate of 3.83 degrees/second.
Damage to this area of the flight recorder foil produced a slight displacement
and rotation with the result that the heading change rate indicated is probably
higher than actual. However, the difference is so small as to be of negligible
importance in this discussion. With the airspeed increasing from 193 to 198
knots at the same time, the bank angle would be about 35 degrees and the increase
m normal load factor due to the turn would be about 0.2 g in a coordinated turn.
- to -
This is roughly in agreement with the mean value of the decreasing acceleration
trace during the same period. Under these conditions the airspeeds would have
been *30 knots or moie above stall with 20 degrees of flaps and 35' knots or
more above with flaps fully retracted.
Starting at time J008; 07 the heading trace began bo record changes the
significance of which cannot be precisely defined. From this point in time,
therefore, the recorded heading changes will be described but the possible
meanings to be ascribed to them, except where the reason for a change is clear,
will be discussed under analysis.
At 1008:07, within six -tenths of a second, a heading change from 250 to
24-3 degrees is recorded; a change rate of 12 degrees/second, amounting to
an instantaneous tripling of the 3*83 degrees/second change rate recorded just
prior to this time. From 1008:08 to 1008:16 the heading trace passes through
the badly wrinkled, abraded and torn portion of the foil, during which time
the heading changed from 245 to 225 degrees. At time 1008:16, the heading trace
emerges from the badly damaged protion of the foil and for the next one and
one-half seconds indicates a much lower rate of heading change than that
immediately preceding entry into the damaged area. The heading trace shows a
momentary cessation of turn at 1008:19 followed by a higher turn rate con-
tinuing until 1008:31, when a sharp reversal in the recorded heading is noted.
Such an abrupt change from a left turn to a right turn is beyond the aircraft's
capability and is an indication of gimbal error in the directional gyro of
the airplane as the angle of left bank approached 90 degrees. From this time
to about 1008:4-2 the recorded headings are similarly in error due to the
high roll and pitch angles of the airplane as it inverted and dove to the
At time 1008:18, 1008:25 and 1008:29 the airspeed and altitude traces
indicate sharp simultaneous increases which were beyond the aircraft's per-
formance capability, and the highest peaks recorded indicate an airspeed of
230 knots and 2,000 feet of altitude at 1008:30. Tne precise significance of
these recordings cannot be completely defined. However, a single static port
low on the left side of the forward fuselage is connected to the airspeed and
altitude sensors of the flight recorder. As a result, nose left sideslip
(relative wind from the right) and high angles of attack cause appreciable
plus errors in the recorded airspeed and altitude.
The median acceleration trace shows a rise from 1.0 to 1.8 g from time
1008:26 to 1008:30 at which time an abrupt change is recorded in a manner in-
dicative of heavy stall buffet, which intensifies and continues until impact.
Traces from the same make and model of flight recorder made during Boeing Company
test flights in a 707-13 IB airplane substantiate this observation.
Immediately following the recorded peak airspeed and altitude, both
parameters drop abruptly, and the airspeed trace records its lowest subsequent
speed of 170 knots at 1008s 37. This rapid peaking and decrease in the airspeed
and altitude traces indicates pronounced sideslip "effects coupled with increased
drag resulting from prolonged heavy buffeting.
The abrupt changes of the airspeed, altitude and acceleration traces pin-
point the time of impact within plus or minus one second. As previously
- 9 -
mentioned the impact time was established at 100$: 4-9 by the Fordham University
An extensive review of the maintenance records of N 7506A was made in
search of information which might have some bearing on the accident. This review
was conducted at the facilities of both American Airlines and The Boeing Company
and, when deemed necessary, maintenance and inspection personnel were interviewed.
The flight logs and the maintenance, inspection and overhaul records were studied
with special attention directed to the flight control systems. Records of major
modifications performed on the airplane were also investigated. One instance
of improper maintenance was found. During intended compliance with an American
Airlines Engineering Change Oder, an outboard bellcrank was erroneously in-
stalled at the inboard bellcrank position of the spoiler controls in the right
wing. No functional or operational check was required or performed upon com-
pletion of the Engineering Change Order. On the following flight, the crew
reported difficulty with un symmetrical spoilers, whereupon maintenance personnel
adjusted the speed brake and spoiler control rods to correct the condition. After
13 additional flights, during which no flight discrepancies were logged concerning
the spoilers, inspectors discovered the error. A correct installation and re-
riggmg were then accomplished on February 25, 1962, with no subsequent complaints
concerning the spoiler system as a result of the intervening three flights prior
to the accident. Except for this instance, all records examined reflected that
the aircraft was continuously maintained in an airworthy condition in accordance
with FAA -approved company policies and procedures.
All four powerplants suffered extensive and similar damage characteristic
of a high-velocity* nose-down impact of the aircraft. The extent of torsional
damage in the four engines indicates that each was operating at approximately
60 percent r.p.m. , which is equivalent to flight idle thrust, at time of impact.
Although detailed examinations were performed on each of the shattered engines,
no evidence of in-flight damage or failure could be found.
The crater made by the aircraft in the bottom of the bay was approximately
130 feet long and $ to 10 feet deep. On impact the wings were fragmented and
the fuselage crushed accord ion -like, breaking into many sections, part of the
horizontal stabiliser and elevator with the tip of the fuselage attached was
the largest piece of structure recovered. The fire, which ensued shortly after
impact, heavily damaged the above-water portions of the airplane structure.
Impact and fire damage was so extensive as to preclude examination of numerous
components of the aircraft which might possibly have yielded important information.
No evidence could be found to indicate that there had been an in-flight fire, an
explosion, structural fatigue, or overload failure.
The cockpit area suffered the most extreme fragmentation of the entire
fuselage, the degree of fragmentation gradually decreasing toward the tail of the
aircraft. The horizontal stabilizer broke loose at FS (fuselage station) 1505,
with the tail cone intact and still attached. The vertical stabilizer tore out
of its structural attaching bulkheads at FS 1440 and FS 1507 at impact.
The landing gear was determined to be in the fully retracted position.
There was no indication of defective treads on any of the tires, nor was there
any evidence of a tire blow-out in any of the wheel wells.
- 10 -
The skin and most of the remaining structure of both wings suffered severe
fragmentation- Examination revealed numerous indications thab all wing flaps
were in the fully retracted position.
"Reconstruction" of the wreckage was made in a hangar for detailed study.
As minute an examination as possible was made of the lateral control system
considering the extensive damage it had sustained. Of the eight control cables
in the right wing, only the right wing down trim cable was missing and all
others were in place. Of the eight cables in the left wing, all were in place
except for the bus cable ABSB, which was missing from WS (wing station) 4-60
outward. In the right wing, most of the inboard aileron was burned away) the
outboard aileron was complete. In the left wing, the inboard aileron was fairly
well intact despite heavy fire damage j the outboard aileron was extensively
damaged by the impact and fire. The aileron lockout mechanism of the right wing
was determined to be in the locked-out position. Major portions of the left
wing lockout mechanism were not recovered, but the actuator screw was found ex-
tended to the locked-out position. Impact damage to three of the four aileron
bus quadrants disclosed that the right inboard aileron was 10 degrees UP and the
left, 10 degrees DOWN at impact. No physical evidence of in-flight failure in
either the right or left aileron system was found. The components of the right
wing spoiler system were damaged extensively during impact. Examination indi-
cated that at impact the right wing inboard spoilers Nos. 5 and 6 were 28 and 31
degrees UP respectively, and the outer panel of the right outboard spoiler was
4-0 degrees UP. The left wing spoiler system sustained severe impact damage
and heavy fire damage, and the spoilers were found in the full DOWN position.
A tear-down Inspection of the eight spoiler actuators and four spoiler control
valves revealed that these components were capable of normal operation prior
to impact. Severe damage to the lateral control system due to impact and fire
precluded the examination of some essential parts. Both cockpit control wheels
were determined to be slightly beyond the position for a full right wing down
Examination of the horizontal stabilizer revealed no evidence of any mal-
function. Measurement of ^he position of the nut on the stabilizer jackscrew
corresponded to a cockpit indication of 2.3 units nose UP trim.
On impact the vertical tail tore completely loose from the fuselage, landed
on its right side in the area of the most intense fire, and the rudder and tab
were almost completely destroyed by fire. The upper portion of the vertical
stabilizer was severely damaged bj impact and fire but the remaining identifiable
parts included the rudder boost unit, the tab damper, and portions of the Q
bellows assembly. The latter was partially consumed or melted but ii?as still
attached to structure; all cables and fittings were in place down to the Q rod
front fitting where the rod had melted a,way. The aft rudder control quadrant
had melted, but the cable attach ends, with attach brackets and bolts still
installed, were found in a mass of solidified metal. However, it was impossible
to differentiate the remaining portions of the control cables from those of "Che
servo cables. The aft fitting of the rod attaching to the rudder boost control
valve was properly bolted and saf etied . No other part of this rod or any portion
of the attaching ratio bellerank, tab control rod, or rudder quadrant control
rod, could be found. The rudder control (compound) bellerank was found with
all the rod ends properly attached. The rudder trim torsion rod was in place
with its upper end pushed up into the sleeve, and thus disengaged, as a result
of impact deformation. The trim drum and gears were recovered and appeared to
- 11 -
be in normal condition. The rudder centering spring mechanism shoved no
evidence of cables climbing out of the pulley grooves.
After removal of the main wreckage to a hangar, the accident site was
combed with hand rakes. U. S. Army personnel, with mine detecting equipment,
later assisted in the search for wreckage. Because of adverse weather con-
ditions and exceptionally high tides, recovery of the wreckage was difficult
and slow. A hydraulic dredge was employed to recover pieces believed to be
imbedded in the muck. This operation was conducted for a period of three to
four weeks during which tune numerous pieces of wreckage were recovered. The
search was continued using a crane with clam-shell digging equipment, resulting
in recovery of additional wreckage. Some of the wreckage recovered was in the
form of metal masses resolidified after having melted. These were given X-ray
examination and in some cases chipped apart for study.
Examination of the airplane communication equipment revealed frequency
selection in the recovered equipment appropriate for the period of flight in
The electrical system was studied for any indications of an electrically
caused fire in flight or the malfunction or failure of any system due to electri-
cal faults. Although the thoroughness of this study was restricted by impact
and fire damage, no evidence was found to indicate that an electrical arc, short
or overload had existed in the electrical system prior to impact. Numerous in-
dications obtained from the wreckage disclosed that electrical energy was present
at impact .
The hydraulic system, also damaged by fire# yielded evidence that it was
operating until time of impact. The previously mentioned positions of the
inboard and outboard spoilers in the right wing, which require both utility
and auxiliary system hydraulic power, is one such indication. Ultraviolet ex-
amination of the face of the rudder system hydraulic pressure gage, located in
the first officer's instrument panel, revealed an outline of the hand at 3>&00
p.s.i. The pressure transmitter, which actuates this gage, checked normal when
The rudder hydraulic pressure control valve, electrically operated, was
determined to be in the electrically deenergized, or 3,000 p.s.i., position
normal for airspeeds below 245 knots. The electrically operated rudder pressure
shutoff valve was also determined to be in the pressure ON, electrically
deenergized position. "When disassembled, it was found that the nickel plating
had partially peeled or flaked away within both of these valves. However, sub-
sequent tests by Boeing indicated that any nickel particles thus released into
the system would not adversely affect the operation of the rudder boost unit
due to a filter at the boost unit inlet.
Rudder damper measurements corresponded to a rudder position of 17.$ degrees
LEFT. The piston of the hydraulic actuator of the rudder power control unit
was extended 3/4 inch from neutral in a direction corresponding to a position of
9 or 10 degrees RIGHT rudder. Due to various factors the heat of the ground fire
•would not tend to produce any changes in this position. This is the rudder
travel which can be produced by 200 pounds of pedal force at 200 knots airspeed
- 12 -
with boost OFF. The actuator control valve spool had moved beyond the position
normal for right rudder movement and had been driven through the rear end of its
housing. Severe fire damage was evident at this point. However, the over-
driven position of the spool was obviously a result of impact forces.
The autopilot disengage switches on the captain's and first officer's control -
wheels were found with the buttons in a depressed position beyond normal travel. ,
A special study was conducted to determine the significance of this finding. Exam-;
mat ion revealed that both operating button plungers had thrust completely through '
the bakelite bottom of the switch. Normal operating pressure on the disengage
button is 3 to 4 pounds. Tests showed that a load of about 90 pounds is required '
for the plunger to break through the bakelite bottom, a force beyond the physical
capability of a pilot's thumb to produce, considering that most of the thumb pres- -
sure would be applied to the switch plate rather than to the button. No marks on
the top of either plastic button were found to indicate they had been struck by .
harder aircraft structure during impact. Tests prior to disassembly of the switches'
and visual examination afterwards showed that the electrical contacts were open in
both switches, corresponding to autopilot disengagement. However, the positions ;
of the contactors were abnormal m a manner consistent with the overdriven con-
dition of the plungers. As a result, the condition of electrical discontinuity is
not indicative of the switch positions immediately prior to impact.
The automatic flight control system was extensively investigated. This system
provides automatic coordinated control of the airplane, and a damper control mode
is available to augment yaw stability when the airplane is controlled manually. 6/
The autopilot is an electronic, electromechanical device that converts small
electrical input signals into mechanical movements of the control surfaces. Sensing,
devices generate signals that are amplified and converted to electrical power used ,
to energize servo motors which actuate control surfaces directly or through hydrau-!
lie actuators. The voltage generated by a sensing device, as it is applied to the
amplifier input, is modulated by other voltages generated by a sensor of control
surface movement. These modulating voltages originate in surface position trans-
mitters and in units that signal the rate at which a surface is moving. The
combined input signal level is in the order of one volt.
The control panel for thus "flight control system was nob recovered from the
wreckage though portions of some of the controls were. The autopilot engage switch
was recovered and was found jammed in a position slightly toward the autopilot
engage posit ion 5 this direction was also that of impact forces.
The control unit contains two rate gyros that sense rotation rates in pitch
and yaw. Transmitters actuated by these gyros generate electrical signals, the
strength and sense of which are governed by angular velocity of the aircraft.
are fed to the input of the main autopilot amplifier, and if the autopilot or yaw
>er is engaged, certain control surface movements will result*
The rate control unit vas recovered wi In the cover of the nousmg collapsed
inward, but no marks to account for this damage could be found. The yaw gyro was
not on its mounts. The pitch and yaw gyros are identical except for the direction
of spin axis, and there was a definite difference in damage tc these gyros. The
6/ The autopilot engage switch located on the autopilot control panel is a
toggle type marked AUTOPILOT -DAKPER and is moved forward to engage the autopilot
for three control channel operation (AUTOPILOT position) and aft for rudder channel.
only operation (DAMPER position). The switch is spring-loaded to the (unmarked)
OFF position. The other two positions are solenoid held in the desired position.
- 13 -
frame holding the yaw gyro had a large section broken from the side and the
missing section was not found. The frame holding the pitch gyro was whole and
The surface servo units are electromechanical devices which convert elec-
trical signals into proportional mechanical forces that adjust the aircraft control
surfaces. Each servo contains an electric motor which is geared to a pulley through
a clutch. In this airplane the pulley drives a cable which is connected to a
tab on each of the primary control surfaces and to the power unit in the rudder
system only. The electrical signal from the amplifier drives the motor and eventu-
ally the flight control surface. As the control surface moves to the desired
position a follow-up autosyn generates a signal in proportion to surface displace-
ment which opposes the original initiating signal and stops the surface at the
Mounted on the end of the servo motor shaft is a rate generator which develops
a voltage in proportion to the motor speed, and in opposition to the initial input
signal. The electrically energized clutch of the servo is engaged when the auto-
pilot (or yaw dasper in case of the rudder servo) position is selected. The motor,
rate generator, follow-up autosyn, clutch and gearing are all enclosed in a cast
aluminum alloy housing. The rate generator and end bearing for the servo motor
shaft are enclosed in a cylindrical projection of the bell housing on the servo
motor assembly. This projection is about 5/8 inch long and about 1-l/S inch in
diameter. A large disc-shaped flange of the end bell is recessed below the end of
the motor case approximately 3/16 inch and is retained in this position by a snap
ring. The gap between the inner diameter of the motor case and outside diameter
of the cylindrical projection is approximately 5/8 inch. One of the wire bundles
in the servo assembly which contains the B wires to the servo motor and rate
generator is wrapped completely around the end bell projection and is snugly held
against it at the corner formed by the projection and the disc.
All of the servo units from N 7506A were recovered and examined with the other
components from the autopilot system. Each servo was visually checked for damage
and none of the servos was found to have been damaged by fire, but all were
corroded from exposure to salt water and marked or broken to some extent. Conti-
nuity and visual checks were made of the wiring and components. Nothing could
oe found in the examination of the aileron and elevator servos to indicate the
existence of a malfunction prior to the accident.
A large portion of the housing of the rudder servo was missing, partially
exposing the servo motor. The rate generator end of the motor was completely
exposed. The continuity check of the rudder servo wiring showed an "open" in the
rate generator circuit. The rate generator was then disassembled but no faults
were found until the protective sleeving covering the wiring to the rate generator
and. motor was removed. The wire with brown insulation and the wire with orange
insulation were found to be severed. The blue wire was holding together with only
one strand. Some of the strand ends of these wires bore the appearance of having
been cut by a sharp edge, some were pinched and flattened, and some were necked
down, as disclosed during microscopic examination- by the Board's metallurgist.
The brown wire connects the output of the rate generator to the input of the
autopilot amplifier. The blue wire connects 18 volts AG to the rate generator
- 14 -
input and the orange wire is the ground or return side of the 18 volts input.
The separations in the wires were adjacent to each other- Th& protect jve sleeving
was then examined and a transverse separation having the appearance of a cut or
slice, with a puncture -like indentation at one point in the sleeving separation,
was seen to be adjacent to the severed wire ends. When repositioning the wires
about the rate generator a series of radial scratches and/or gouges was noted
on the end bell of 'the servo motor under and in line with the wire bundle damage.
The sleeving separation extended approximately three -fourths of the way around its
circumference, one end being below center on the extreme outer peripnery of the
bundle as wrapped around the end bell projection. The separation extended from
this point up around the top of the bundle and down the side adjacent to the pro-
jection approximately one-third of the distance to the extreme inside diameter of
the bundle, as wrapped around the projection. -Tust below this end of the separation
there are three indentations m the sleevjng, two of them connected by a linear
depression in line with the separation. The lower two of these indentations are
on the sleeving surface normally in contact with the surface of the projection.
Approximately 1/2 inch from the separation area and on the surface of the sleeving ;
adjacent to the surface of the projection, there are several additional indent at ioni
and punctures. It was also observed that the wire and sleeve damage and the '
scratches on the end bell were in an area somewhat protected by the end bell pro-
jection and the projection of the motor case above the end bell disc. No marks
could be found on this assembly in the area of the wire damage which would indicate
that the wire bundle had been struck by some object or otherwise damaged during
the break-up. On the end of the servo motor case at the point nearest the clutch
housing, the outer surface of the case is flattened and scratched by interference
with some other object.
To determine, if possible, whether the damage found on the rudder servo was
unique, the aileron and elevator servos were reexamined. The end bell surface of
the elevator servo motor was found marked and, although corroded, the surface of
the end bell also appeared to have been scratched or marked. Eight spare servo
unit motors from the American Airlines stock were then examined, and six of
these had the same tjype of scratching or gouging as found on the rudder servo from
N 7506A. Some of them also had similar indentations or imprints on the sleeving
enclosing the wires.
In each of the cases examined the marks on the end bell and sleeving were
observed to be adjacent to each other and re was also noted that this damage
occurred at the same radial position on the end of each motor. In each case \
the sleeving was marked or indented, the scratch marks appeared op the end bell.
One of the scratched units from the American Airlines stock still bore the
manufacturer's seal indicating that it had never been disassemDied since last
leaving the factory.
As a result of these findings, inspection of servo units was made on the
production line at the manufacturer's plant. Board 5_nvesbigators enlisted the
aid of the FAA manufacturing inspectors who found six unsatisfactory units.
Marks, indentations and electrical wire damage within the slel-ving were found
which was similar to the damage previously mentioned. FAA inspectors determined
that this damage had occurred as a result of improper use of tweezers when tying
the wire bundles to the motor housing. Additional units were found to have marks
and damaged protective sleeves, but no wire damage within the sleeves.
- 15 -
At the request of Board investigators, the damaged rudder servo unit from
N 7506A was subjected to a searching examination and analysis by the manufacturer
at his plant. In a report of this examination, which was later submitted to the
Board, the manufacturer concluded that it would be highly improbable for the unit
to pass the electrical requirement of the final assembled servo if a lead or leads
within the protective sleeving were severed during assembly. The manufacturer
further concluded that his examination indicated that the damage could have been
the result of flying fragment damage to the sleeving and leads at the moment of
Discovery of damage to the wire leads to the autopilot rudder servo from
N 7506A dictated studies to determine what effects such damage could have had on
the performance of the airplane. The Boeing Company was requested to perform the
As a result, bench tests were first conducted to determine what autopilot
system degradation or malfunctions could occur. These tests showed that a hot wire
failure of the 18-volt excitation lead, i.e. blue or orange, making contact with
the brown signal lead, produced a "yaw damper hard -over" 7/to the left or right,
depending on the particular connections used. They also showed that loss of
excitation voltage as a result of the severed 18-volt lead produced only insuf-
ficient servo damping.
Flights were also conducted in an attempt to duplicate the bench test
malfunctions while simulating approximately the flight conditions of N 7506A. A
Boeing 707-1313 was used in the flight tests. Duplication of the crossed wires
malfunction in one maneuver, starting from a 30-degree banked turn to the left at
constant altitude produced in eight seconds a left rudder deflection of 7 degrees
at 210 knots IAS, causing the airplane to sideslip and to roll to the left.
Although the hard-over signal was continuously applied throughout the maneuver and
recovery action was delayed for four seconds, sufficient aileron control was
available to stop the roll at 56 degrees in one and one -half seconds and then , to
level the wing. Measurements showed that a rudder pedal force of 75 pounds §/ was
required to move the rudder back to neutral against the hard-over servo force.
Duplication of the open wire malfunction produced a small amplitude oscillation
which was hardly perceptible in the response of the airplane. These flight tests
were conducted in conditions of 1.0 g flight loads-.
7/ Previous failure analysis of the autopilot system had resulted in the con-
clusion that electrical/electronic failures can occur which would result in a
maximum servo force being exerted continuously to drive the control surface against
its opposing air forces. The condition of unwanted maximum rudder servo force
Deing applied constitutes what is commonly termed "yaw damper or autopilot hard -over."
FAA-type certification requires that application of such a malfunction will not
produce hazardous airplane attitudes or loads, assuming the pilot will initiate
corrective action three seconds after the malfunction manifests itself. The rudder
servo motor is capable of commanding only 7° to 8° rudder deflection in the speed
range of 200 to 210 knots IAS.
8/ The slippage torque of the servo motor used in this test was approximately
15 percent below maximum allowable. A force of 90 to 110 pounds would be necessary
to overcome a yaw damper hard-over with a servo motor developing maximum torque.
Boeing certification data shows that with the maximum torque value of 118 inch-
pounds a rudder deflection of 8 degrees is possible.
- 16 -
Other studies' and flight tests were made by the Boeing Company in an effort »
to pursue various leads in the investigation. One of these tests involved known
incidents of the binding of a primary flight control system due to failure of a t
cable pressure seal which prompted tests to obtain more data on the el'fects of «
this type of malfunction. These tests showed that, though increased pressure on -j
the controls was required, continued control movement to overcome the binding was ;
well within pilot capability. Other Boeing studies concerned the possibilities
during the flight of N 7506A of: a rudder boost hard -over; jammed control wheel 5
jammed outboard aileron; jammed spoilers; stall during wing flap retraction; en- ,
gine failure; and pilot inattention, distraction or incapacitation. As a result
of these studies Boeing concluded that no single one of the airplane malfunctions \
considered should cause loss of control, as evident in this accident. \
The carrier conducted a study of the flight recorder traces based on energy r
gain and /loss throughout the flight. By consideration of total head 2/ and specific
energy iQ' histories derived from the altitude and airspeed traces, and consideration*
of the median acceleration trace, "actual" altitude, speed, lift coefficient and ^
sideslip histories were deduced. However, since energy is greatly affected by I
engine thrust, three variations of total thrust during the critical phases of the *
flight, from time 1008: H to 1008:30, were considered. These are: continuing full!
thrust, a 25 percent reduction in thrust, and a 50 percent reduction in thrust. I
These separate conditions produce significant differences in the sideslip histories!
With no reduction in power the altitude and airspeed trace variations correspond tol
a sideslip history of low magnitude wavering through zero and reaching a maximum off
3 degrees nose left slip at 1008:30. The cases of power reduction result in a shift
of the sideslip history to nose left with greater resultant slip angles, the maximui
being approximately 7 degrees nose left for the 50 percent power reduction at time 1
1008:30. Insofar as the energy analysis alone is concerned the two cases of power '
reduction apply equally to either symmetric or asymmetric engine thrust. The lift ?
coefficient histories change with the assumed variations of total engine thrust anfl,
reach values of approximately 1,00, 1.06 and 1.11 at 1008:30 for the 100 percent, \
75 percent and 50 percent thrust conditions respectively. From this analysis the *
carrier then reasoned "that it is most improbable that the significant disturbance *
occurred on the yaw axis, either through thrust asymmetry or through rudder action.'
Further, the carrier stated "to the contrary, we conclude that an initial and
critical disturbance on the roll axis through the ailerons is a much more likely J
possibility ..." The lift coefficient from the carrier's energy analysis
corresponding to this belief of the carrier is approximately 1.00 at time 1008:30 ;
when the flight recorder acceleration trace (Attachment 2) reached its first
abnormally high peak. 1
A program of flight tests known as "Project RAGE" was originated by the Federi
Aviation Agency in an effort to shed light on the cause of the accident. Organi- "
zations participating in this effort to varying degrees included the National 5
Aeronautics and Space Administration, American Airlines, Boeing and the Civil ^
Aeronautics Board. All tests were flown in an FAA -owned Boeing 720 with FAA pilots 4
controlling all flights and performing the maneuvers in all instances: . These tests'
had three main objectives including: provision of flight recorder traces for -
comparison with those made by Flight One by attempting to simulate possible flight [
conditions of N 7506A; measurement of the response of the airplane to various
pilot opposed malfunctions , particularly of the rudder control system; and measure-
ment of the effects of slips and skids by means of extensive test instrumentation -
2/ Total head is equal to the sum of the dynamic pressure measured at the pit*
head and the static pressure measured at the static port.
10/ Specific energy is equal to the sum of the potential and kinetic energy ;
divided by the weight of the airplane.
- 17 -
to make possible a more definitive study of the flight recorder traces from
Flight One. As a result of this program the FAA concluded, in a January 1963
draft of the Project RACE Report, that the "data from autopilot 11/ hard -over
rudder tests do not appear to resemble trace characteristics of the accident
data. Data from Project RACE tests of pilot hard -over rudder, however, ap-
pear somewhat more severe than the CAB read -out of the accident flight record-
er tape." This draft report stated also, "It was found that the automatic
pilotil/ system could not, within the limitations of its force authority,
displace the rudder control sufficiently to develop sideslip angles to the
extent that would cause uncontrollable lateral roll.
"During our tests simulating American Airlines Flight 1 configuration it
was found that the rudder boost system did have this capability. Should the
rudder boost system command full rudder, within the limitations of hydraulic
pressure and aerodynamic resistance, the resulting sideslip angle would cause
lateral rolling that could only be arrested by reducing rudder boost pressure
or assisting lateral control by deploying symmetrical speed brake handle or
Project RACE provided the Board with much information that will prove
helpful in future accident investigations. Insofar as investigation of this
particular accident is concerned, the Project RACE test data are considered
valuable principally as corroboration of more applicable Boeing flight test
data in some respects. However, the Project RACE tests were made in an air-
plane having lower thrust engines, a shorter fuselage, lower moments of
inertia about the J and Z axes, and lower gross weights than necessary for
close simulation of the conditions of Flight One. The lower gross weights
necessitated power reduction for comparable performance, which resulted in
decreased thrust asymmetry with one engine idling, or too much thrust reduc-
tion to produce asymmetry equal to the loss of one outboard engine during
Flight One. The special tests by Boeing in a 707-131B were closer in these
respects. In addition, the Boeing tests to simulate yaw damper and rudder
power control malfunctions included more realistic rudder deflect ion -time
histories than those of Project RACE. Both the Boeing and the Project RACE
malfunction tests were essentially "1 g" maneuvers which did not approach the
high vertical accelerations and lift coefficients experienced by Flight One.
ANALYSIS AND CONCLUSIONS
Throughout the investigation numerous possibilities as to the cause of
the accident were considered and the merits of each were carefully examined.
With the evidence that was amassed, all possibilities were narrowed down to
the following areas which will be discussed in this report: physical incapac-
itation of the crew; loss of engine power ; loss of lateral control; malfunction
of the rudder boost system; and malfunction of the rudder servo unit. However,
prior to detailed discussion of the causal areas there are several subjects of
pertinent interest which must be treated.
It is important to keep in mind that, except where otherwise specified,
the stall speeds referred to in this report apply only to coordinated flight
11/ Equally applicable to yaw damper.
- 18 -
conditions at the approximate CG position of Flight One and, that stall speeds
for high sideslip angles and extended spoilers are higher. It is also impor-
tant bo note that initial stall baffet is caused by separation of the airflow
on the inboard portions of the wings. With increasing angles of attack the
stalled area spreads outward until the wing is completely stalled. Initial
buffet may occur at even high margins above stall speed due to- sudden shock
exerted upon the airflow, such as turbulence or a rapid aileron control appli-
cation. If the speed margin prior to such initial buffet is large and the
initiating influence is of short duration, the buffet will cease- on removal of
the influence. Plowever, if the aircraft was operating very close to the initial
stall buffet speed immediately prior to the aggravating influence, the stall may
well persist after removal of the input. Under this condition, an appreciable
decrease in angle of 'attack is required to restore laminar flow. The initial
buffet felt by the pilot results from the start of a stall in the inboard
sections of the wing, and control surfaces in these areas become less effective.
With the outboard ailerons locked out, as is the case when the flaps are fully
retracted, lateral control is then reduced markedly under partially stalled
conditions, reducing further as the stall progresses. However, this lateral
control still is sufficient to comply with the CAR 4-b stall requirements.
Boeing 707 type aircraft are equipped with a stall warning device in the
form of a "stick shaker" which vibrates the control column to warn the pilot
of an impending stall. With flaps extended 20 degrees this warning device
actuates at speeds seven or more knots higher than noticeable buffet, with
little change in the margin during sideslips up to 5 degrees. However, with
flaps up and zero sideslip, stick shaker actuation and initial buffet are at
the same speed. With 5 degrees sideslip, there is approximately a 5 knots
differential between the speeds at which the stick shaker actuates and those
at which initial buffet occurs-. As shown by the pilot's and copilot's air-
speed indicators, the sense of this differential for a given sideslip direc-
tion is dependent on -&he particular airspeed indicator referred to. For nose
left sideslip, the pilot's airspeed indicator reads high with the result that
stick shaker actuation occurs at lower than stall buffet speed as shown by his
airspeed indicator. Nevertheless, with either flaps UP or at 20 degrees the
device actuates at speeds 18 to 20 knots above the CAR stall speeds. Later,
these relationships will be referred to during discussion of the causal areas.
It must also oe borne in mind "&hat swept -wing airplanes' are subject to a
more pronounced roll- yaw coupling than straight -wing aircraft. This roll due
to yaw t*as referred to as "dihedral effect" on straight -wing airplanes* When
a swept -wing airplane with dihedral yaws, not only is the advancing wing at a
higher angle of attack but it also presents a greater span to the airstream.
.Also, the retreating wing is less effective due to the change in airflow to a
more spanwise direction. The lift differential developed by the swept wings
is therefore higher and produces a greater rolling moment than would be ex-
perienced with a straight-wing airplane under similar conditions. It follows
wherefore that roll due to yaw input of the rudder is much more pronounced on
swept -wing than on straight -wing aircraft.
Physical incapacitation of the crew ? Unrecoverable body tissue vital
to complete medical evaluation, resulted in a lack of conclusive positive or
- 19 -
negative proof of physical incapacitation. However, toxicological studies
were conclusive in ruling out incapacitation due to toxic gases, alcohol and
drugsj and the crew's medical histories also disclosed no reason to suspect
Flight One's last radio transmission at 1008:09 revealed no sign of crew
incapacitation. Though not conclusive, an indication that both pilots were
alert and conscious at impact was the downward bending of each right rudder
pedal, evidence that both were applying pressure to these controls at the time
of impact. The fact tnat the control wheels were found calling for full right
wing down is also indicative that at least one of the two pilots was still at-
tempting recovery action at impact. It is apparent, then, that any incapaci-
tation of the crew would have occurred only between 1008:09 and some undetermi-
nable number of seconds before impact, an interval of less than 4$ seconds.
The flight recorder indicates the first deviation from normal climbout to
start ac 1008:12 and that flight conditions at about 1008:30 were beyond the
possibility of successful recovery action. The 21-second interval between
1008:09 and 1008:30 is, therefore, the most logical interval of time to be
considered wherein crew incapacitation might have occurred.
The possibility of both pilots becoming physically incapacitated simul-
taneously is so remote as to be eliminated from any consideration whatsoever.
The history of incidents involving crew incapacitation during flight has
yielded little information to date concerning the effects of such incapacita-
tion on the controllability of an airplane. Involuntary control forces that
might be applied could vary from a negligible to a substantial force. However,
incidents of operator disablement while driving motor vehicles has indicated
that severe pain usually causes the driver bo double over or slump forward, and
that disablement due to a heart attack is not usually so severe that the driver
cannot pull off the road and stop.
It is reasonable to believe that during the departure from Idlewild either
pilot was in a position to immediately assume control of the airplane in the
event the other was disabled. Furthermore, the second officer and the flight
engineer were available to assist in the restraint of an unwanted control
input by a disabled pilot.
Within the period of time in question, 1008:12 to 1008:30, there were
18 seconds in which the remaining crew members could have restored control
of the airplane had there occurred an incapacitation of one of the pilots
during the early part of this interval. It appears highly improbable that any
control input during the period 1008:12 to 1008:30 would be of such magnitude
and duration as to prevent correction by the other crew members within the
time indicated. In view of the foregoing factors, the Board considers it un-
likely that physical incapacitation of either the captain or first officer
was a causative or contributing factor in this accident.
Loss of engine power : Examination of the engines disclosed no evidence
of an abnormality which would affect their operation.
Witness testimony relative to cessation of smoke trails from the engines
tends to support one analysis of the flight recorder data which indicated a
- 20 -
power decrease near the apex "of the climb. Observations of -numerous takeoffiEs
of ^et aircraft using the same kind of engines as N 7506A indicate that visual
impressions of smoke trails are unreliable due to many factors, including:
viewing angles, changing flight maneuvers, relative angle and amount of sun-
light, and variances in the amounts of smoke emitted by individual engines.
Therefore, the true significance of witness observations cannot be properly
assessed. However, there very probably was a power reduction in the late
stages of American One's flight.
American Airlines energy analysis and flight tests by Boeing indicate that
maximum power must have continued until approximately 1008 rl4 in order to pro-
duce airplane performance consistent with the flight recorder traces. The
energy analysis indicates further that from 1008:14 to lOQ&t'ZS the thrust his-
tory could have varied anywhere from continuation of maximum power to a 50
percent reduction. The energy analysis does not 3 of course, provide any indi-
cation as to whether any possible power decrease considered was intentional or
The flight tests indicated that a total loss of power from the left out-
board engine, the most critical loss to be considered, would not have presented
a critical problem in maintaining control of the airplane. The simultaneous
loss of two engines on one side is believed so improbable that it does not
The Board therefore concludes that a loss of engine power was not an
initiating 01 contributory factor in this accident. However, such a conclusion
does not eliminate from consideration the probability of an intentional power
reduction by the crew in an effort to macaitain control of "the airplane.
Malfunction of lateral control system ; No positive indication of any mal-
function in tne lateral control system was found during detailed -examination of
the wreckage. However, many critical parts were either unrecovered or melted
down* with the result that there could have been a malfunction in one of these
One area of possicle discrepancy found during examination of the wreckage
was that marks made on tne aileron cable bjs quadrants at impact corresponded
to the right inboard aileron being about 10 degrees UP at the time, with other
impact damage indicating that the control wheels were beyond the full right
wing down position, the right inboard spoilers about 28 and 31 degrees W and
the outboard section of the right outboard spoiler about 4-0 degrees IF. Since
the airspeed at impact was about 200 knots, as indicated by the flight recorder,
normal operation of the lateral control 'system "with wheels at full throw would
have produced 20 degrees UP right inboard aileron, and 40 degrees IIP right, inboar
spoiler, without use of speed brakes to augment, lateral control. This dis-
crepancy tends to lend credence to the possibility of some malfunction in the
lateral control system.
A study made by Boeing indicates that if an outboard aileron is jammed, for
example by ice on the balance panels, the action of the lockout mechanism on the
connecting quadrant during flap retraction from 20 degrees to aero degrees, can
actuate the other aileron surfaces through the bus cables. If the left outboard
- 21 -
aileron is more than two degrees up when jammed, the resultant left roll from
the flap -driven aileron surfaces cannot be overcome by control wheel effort alone.
The study indicates further that as the flaps retract through the range of 13
to 9 degrees, the load in the link rod of the binding aileron exceeds its
design strength and fails, after which the remainder of the system is freed,
permitting nearly normal control of the aircraft.
However, there appear to be additional possibilities in connection with a
jammed aileron which could be pertinent to this accident. One of these is that
deflections or failure at another point, unanticipated in the Boeing analysis
and not disclosed by the ground tests on which it was based, could result in
full flap retraction without failure of the link rod. This could result in at
least three of the four ailerons being held in deflected positions, the amount
of deflection depending on the jamming condition, aerodynamic loads, cable
stretch, and other variables. The spoilers would still remain operable through
the cable system from the control wheels.
Another possicility, one suggested by the control position discrepancy at
impact, is that although failure of the link rod is accepted, the captain and
first officer could reasonably be expected to apply lateral control effort to
the limit of their physical capabilities prior to the link failure. The result-
ing force would load the aileron control system from the control wheels through
mechanical linkage to the tabs on the inboard ailerons and to the spoiler control
valves. As a result, abnormal pilot input failures at certain points in the
system appear possible, such as deformation of the sleeve between the control
wheel and the control column, or the terminal at the bottom of the control
column. Such deformations could result in less than normal lateral control
being available after the flaps are fully retracted.
If, as discussed earlier in this report, the flaps were retracted from
20 to zero degrees between times 1007:57 and 1008:09, the possible dog leg
in the flight recorder heading trace as the result of gimbal error at high bank
angles between times 1008:07 and 1008:17 is in general agreement with a left roll
produced by binding of the left outboard aileron. If flap retraction did not
cause failure of the outboard aileron link rod, or if abnormal pilot effort
caused control system deformations, the left roll could continue despite maxi-
mum opposing control wheel effort. Rapid application of right rudder could
then be expected. This should yaw the airplane nose right and roll it out of
the bank, since the flight recorder acceleration trace indicates no probability
of the wing being stalled at this time. However, the flight recorder traces do
not indicate any right yaw until about time 1008:19, and this in only a small
fraction of that which could be produced by rudder effort.
Using the actual speeds from the energy analysis and median values from
the flight recorder normal acceleration trace, as indicated in Attachment 2,
lift coefficient histories were determined. Comparison of these at time 1008:30
with the lift coefficients for heavy stall buffet as determined by Boeing tests
discloses agreement only for the 50 percent thrust condition. This implies the
start of a nose left sideslip at time 1008:12. The only apparent logical way in
which a nose left sideslip could have started at this time in a manner necessary
to satisfy the energy analysis, would be the loss of power from the Nos. 1 and 2
engines as a result of the unwanted roll. However, no reason for such power loss
- 22 -
can be seen without assuming other independent failures. As a result, these
types of lateral control failure do not appear to be a causal factor.
There has been at least one reported instance involving another 707 which
experienced difficulty in rolling out of a 30-degree banked turn. Although at
did not involve a lateral control malfunction, it appears pertinent for discus-
sion at this point. After takeoff, while flying in extremely rough air, the
airplane levelled off at 2,800 feet, a 120-degree turn to the right was started,
flap retraction from 20 to zero degrees initiated at \ r 2 / 30-, and power reduced.
When attempting to roll out on the new heading, left aileron was applied rapidly,
but the low wing failed to come up. Left rudder was then applied to bring the
wing up, but the bank and turn continued and the descent rate increased. Since
the airplane was not responding to normal recovery actions, the pilot applied
additional power, reversed aileron control to the right with the turn, and pushed
the nose down. Recovery was then effected. The crew reported that there was no
evidence of a stall at any time jn this sequence of events. Although the crew
did not recognize stall buffet, probably due to the very high turbulence, it
appears certain that the wing was at high angles of attack and in the buffet
region where the flaps -up lateral control becomes less effective and can be in-
sufficient to raise a low wing. Although lateral control deterioration probably
occurred on Flight One after approximately time 1008:28 and added to the diffi-
culty, the flight recorder braces indicate that abnormal conditions started at
least 16 seconds earlier due to other reasons.
One last and important aspect of the lateral control question not yet
discussed lies in the finding that the flaps had been left in the retracted
position as indicated by the physical evidence after impact. Had the crew
recognized their difficulty as one of lateral control it would be reasonable to
expect that they would have extended the flaps in order to regain use of the
outboard ailerons. Two other recovery methods were also available: asymmetric
power and rudder control. Considering the recovery methods available, as applied
solely to a lateral control malfunction, it does not appear likely that such a
The Board therefore believes it unlikely that a malfunction in the lateral
control system was a causative factor m this accident.
Malfunction of rudder boost system : In this airplane the rudder control
system incorporates a boost unit which is in operation throughout normal flight.
Pilot forces applied to the rudder pedals operate control cables running to the
aft end of the fuselage, and up into the vertical fin structure where they connect
to the aft rudder quadrant. Gables extending upward from the rudder servo unit
also connect to this quadrant. An actuating rod extending rearward from this
quadrant is connected to the ratio bellcrank, which in turn is connected both
to the control valve of the power unit by means of one rod; and to the rudder
tab by means of another rod, a compound bellcrank in the rudder leading edgo, and
finally, a rod bolted to the tab horn. A Q-spring assembly supplies modifying
forces to the rudder control system, maiuily to provide artificial feel vnen the
rudder is deflected more than 17 degrees. The Q-spring system connects to the
primary rudder system at the compound bellcrank in the rudder leading edge. When
a force is applied to the rudder pedal the resultant rotation of the ratio bell-
crank actuates both the tab and the power unit control valve causing both
- 3 -
aerodynamic forces and forces from thL. power unit to move the rudder in the
desired direction. The power unit frame is connected directly to the rudder
by a pivot bolt, and the piston rod of the actuator of this unit is connected
to stationary fin structure by means of another pivot bolt. This results in
displacement of the power unit case with movement of the rudder. This movement
of the case provides follow-up action which centers the control valve spool when
the rudder reaches the desired deflection. The rudder power system normally
receives hydraulic pressure from one auxiliary hydraulic pump and during takeoff
and climb American Airlines requires both auxiliary pumps to be ON to supply '
pressure to the rudder power system- At speeds below approximately 250 knots the
system pressure is at 3,000 p.s.i. As airspeeds increase through 250 knots an
airspeed switch actuates the rudder pressure control valve, reducing the hydraulic
pressure to 1,000 p.s.i. The power control unit in N 7506A could be deactivated
by operation of a guarded toggle rudder boost switch located in the right rear
corner of the overhead panel in the cockpit. Since the accident, the carrier has
relocated this switch to the approximate center of the overhead panel, making it
more readily accessible to both the captain and the first officer. The forward
position of the switch is y ON and its aft position is OFF. The guard on the toggle
switch protects it in the ON position and must be raised to actuate it to the OFF
Damage to various components of the rudder system gave conflicting evidence
of rudder position at impact. However, study of various factors indicates that
the most reliable evidence of rudder position was that indicative of 9 to 10
degrees right rudder deflection. The impact deformation to the right rudder pedal
assemblies, distinctly different from that to the left rudder pedals, was in-
dicative of both the captain and the first officer applying right rudder pressure
at time of impact. The fact that the right inboard and outboard spoilers were
found extended is indicative of both auxiliary and utility hydraulic pressure to
time of impact. This is an indication that the hydraulic quantity was sufficient
to supply hydraulic pressures for normal operation of all systems, including the
rudder power system. The rudder system hydraulic pressure gage indication of
3j 800 p.s.i, is above the normal operating range. This position during impact
breakup could have resulted from the needle being displaced from the normal
3,000 p.s.i. range due to hydraulickmg effects at impact. It is also possible
that immediately prior to impact the needle could have been at zero due to previous
actuation of the rudder boost switch to the OFF position, and that distortion of
the gage at impact resulted in the hand moving counterclockwise from zero to 3* #00.
The rudder boost switch guard, the handle and the switch mechanism were missing
and it was not possible to determine if this switch was ON or OFF prior to impact.
Any failure in the control valve link rod, the ratio bellcrank, or structure
supporting the bellcrank; or disconnect of either the bolt attaching the rod to
the bellcrank or the pivot bolt for the ratio bellcrank, would prevent normal
application of both control input and follow up action to the control valve.
The bolt connecting the actuating rod to the control valve was found still m
place. However, as stated previously, the above mentioned parts were not recovered;
therefore, no determination could be made m regard to continuity in this area.
The possibility of a disconnect of the bolt attaching the ratio bellcrank to
the forward end of the valve actuating rod was given considerable attention during
the investigation. It must be noted that this bolt has a countersunk head and
is installed head down to avoid interference with a stiff, flexible hydraulic hose
connecting to the power unit case. This hose passes directly under the bolt head
with approximately l/4.-inch clearance. It is pertinent to note that the hose
- 24 - i
axis is essentially parallel to control input movement of tne bolt in question. j
Therefore, if the securing nut, normally safetied by a cotter pLn, were missing
the bolt could drop down and contact the hose where it would ride back and forth -
with subsequent movement of the controls. It is possible that the bolt could
drop out entirely free of the bellcrank and rod en<$ ? depending upon the particular
aircraft installation. Tf the sharp-edged bolt head should come to rest on the
hose, the resultant rubbing action could cause wear and fouling of the bolt with
the hose, either restricting control movement and /or rupture of the hydraulic
hose. A worldwide inspection campaign of 707 type aircraft required by the FAA
since this accident disclosed that in all aircraft this bolt wis properly installer
A study of the results from Boeing and Project RACE tests, in conjunction '
with the flight recorder traces for Flight One, indicate roll effects from side-
slips which could possibly result from a malfunction in the rudder boost system
caused by any of the control valve disconnects mentioned above. Control valve
unportmg which may result from such disconnects could be sufficient to cause
full hydraulic flow rate to the power cylinder, or it could be at a lesser rate
due to the throttling effect of a small uncentering of the rudder control valve.
These two variations will be discussed separately in the following paragraphs .
Considering first the case of a full hydraulic flow rate to the power
cylinder (maximum rate hard-over) starting at about time 1008:12, the variations
of indicated altitude and airspeed shown in Attachment 2 do not correspond to
the high sideslip angles which can be predicted as a result of full rudder dis-
placement. Tne Boeing test data show that maximum rudder deflection would
probably occur in less than two seconds with maximum rate hard -over producing
extremely violent airplane response. At the probable high rate of rudder deflec-
tion, any attempt to correct with normal lateral control alone would not stop
the resultant roll and sideslip.
In less than four seconds the sideslip would build up to about 14- degrees
which is two times the maximum sideslip reasonably deducible from the flight
recorder traces and at a rate of sideslip increase about 8 times greater.
The use also of 20 degrees of speed brakes, with only one second delay in
starting the recovery attempt, would produce sufficient control to stop the roll, ,
but not sufficient to decrease the bank angle. However, approximately the same
sideslip angle and sideslip rate would remain, which again is not in agreement
with the flight recorder traces.
The use of lateral control and maximum asymmetric thrust, with only one
second delay m applying both, would counteract the roll and sideslip, but the
maximum slip angle and rate would still be much greater than indicated by the
flight recorder traces. From a practical viewpoint it appears highly unreasonable
i-o assume that the pilot would accomplish this sequence of corrective actions
in the one -second time interval. Any additional delay would make the disparity
even greater .
As a result it is concluded that this accident could not have been initiated
by a maximum rate rudder hard -over.
Considering secondly the case of a small uncentering of the control valve,
the flow rate could conceivably be throttled sufficiently to reduce rudder
- 25 -
deflection to produce sideslip effects grossly consistent with the angles and
rates indicated by the flight recorder traces from time 1008:12 to 1008:26.
This would imply application of asymmetric thrust after a delay of about six
seconds, as indicated by the cessation in sideslip increase from tame 1008 r 19
to 1008:22 in the American Airlines analysis for 50 percent thrust reduction.
Such a delay in applying thrust asymmetry appears more reasonable than
any lesser time delay, since first attempts to take corrective action with
the control wheel are more instinctive. The increasing sideslip after 100$: 22
could then result from the increasing rudder displacement caused by the unported
control valve, and after about time 1003:28 with decreased lateral control
effect iv it y as the wing angle of attack increased. With maximum aileron effort
being applied and nose high stabilizer trim corresponding to that at crash impact,
it appears possible that the pitch-up indicated by the acceleration trace could
have resulted from an entirely unintentional small change of the elevator control
force as a direct result of the high aileron control forces being applied, as the
pilot concentrated with great physical effort on lateral recovery. Carrying this
possible sequence still further, boost disconnect at about 1008:32 would also
tend to rehuJt m the nose right sideslip indicated by the flight recorder air-
speed trice due to the cessation of the rudder input with power asymmetry and
opposite aileron still applied. Cutoff of the remaining two engines shortly after-
ward still leaves time for the reduced rpm indicated by the torsional damage to
all four engines at crash impact.
The Board therefore concludes that a throttled rudder control valve mal-
function could have been the initiating abnormality which resulted in the accident.
Malfunction of the rudder servo unit : The point at which the rudder servo
connects to the rudder control system is mentioned in the preceding section
concerning the rudder boost system. The method in which the rudder servo operates
is discussed in detail in the Investigation section. However, it is pertinent to
reiterate some of the salient features of operation at this point. The servo
motor drives a cable pulley through a clutch which limits the force authority of
the servo. Since the cables from this pulley attach into the rudder system at the
aft quadrant, control forces from the servo produce exactly the same effects as
equal cable loads from the rudder pedals. However, the clutch in the servo unit
is so designed as to permit overpowering of the servo by application of pilot
forces to the rudder pedals m the event of any probable malfunction, including
false electrical signals. The American Airlines 707 checklist specifies engage-
ment of the yaw damper, of which the rudder servo is a component, shortly after
takeoff. The heading trace shown in Attachment 2 changes from a wavering line
to a straight Ixne at 1007:38, suggesting yaw damper engagement at this instant.
The investigation disclosed only one instance of unairworthiness of N 7506A
at the time of the accident; that of the wire damage to the rudder servo unit. As
previously indicated, the nature and protected location of the wire damage precludes
the possibility of such damage having occurred at impact. The additional fact
that numerous servo units were found on the assembly line with similar damage and
markings is considered to be conclusive evidence that the damage to the rudder
servo unit of N 7506A was initiated by assembly or maintenance operations. Follow-
ing the original damage it is believed that tensile strain in the securing of the
wire bundle caused wires that were damaged but not completely severed to be necked
down and weakened to the extent that vibration and other disturbances over a period
- 26 -
of tune caused their final separation. The severed wire ends disclosed no
evidence of melting or deposits characteristic of arcing; however, the low
voltages and high impedances involved would not produce an arc of 1 sufficient
intensity to create such evidence.
Flight tests have demonstrated that separation of the wires without sb orbing
results only in a loss of damping which is hardly perceptible to the crew in the
speed range under consideration. The final wire separations therefore could have
occurred during Flight One or prior thereto. However, a yaw damper hard-over
occurs when there is shorting between the proper ends of the damaged rate generator
leads- By reference to Attachment 2, this appears likely to have occurred at
time 1008:12, where the recorded altitude and airspeed indicate the start of an
abnormality. Shorting at this time could have been brought, about by the inherent
tendency of severed leads to untwist from a twisted bundle, as well as by the
loosening of the loop around the rate generator case as a result of the wire
separations which makes shifting due to vibratory loads much more likely.
It has been established that shorted rate generator leads can produce a
maximum rudder deflection of 8 degrees in 8 seconds, which in turn results
in a roll to 56 degrees in 5-1/2 seconds, starting from a 30-degree bank at 210
knots IAS. It is significant to note that maximum aileron recovery action
during flight tests was started 1-1/2 seconds prior to the airplane reaching 56
degrees. During this 1-1/2 second interval, the roll increased 13 degrees. Test
data establishing the foregoing was based on flight conditions at essentially 1
g acceleration loads. Furthermore, the tests are obviously planned maneuvers
under which conditions the pilot is not confronted with the necessity of analyzing
the malfunction, deciding what corrective action he will take, and experimenting
to produce the desired results. In addition, when considering the operating
conditions of Flight One, there were several distracting influences such as
departure procedures, radio communications, flap retraction, turbulence, lack of
visual horizon reference ahead due to the nose-high attitude of the aircraft, and
the excellent weather conditions which would decrease frequency of reference to the
attitude instruments. As a consequence it is unreasonable to assume that under
the operating conditions of Flight One at this time the pilot, confronted with an
unexpected roll, would start corrective action as soon and to the extent charac-
teristic of planned flight tests.
The above is borne out by recorded instances of yaw damper malfunction or mis-
management. In all instances the crew was late in recognizing the yaw damper as
being the source of the problem and were slow in initiating corrective action. Jn
some cases, even after initiation of corrective action the dangerously steep banked
attitudes increased and persisted well beyond flight test values before recovery
was effected. In some instances of yaw damper mismanagement the crew never
properly analyzed the difficulty and the flights were completed after application
of additional lateral control such as use of speed brakes, flaps extension, etc.
There are some instances wherein the crew took advantage of additional lateral
control capabilities, recovered to level flight, analyzed the difficulty, and then
disengaged the offending yaw damper.
Returning to time 1008:12, the beginning of the nose left yaw damper hard-
over, it appears from the flight recorder traces that the airplane was in about
a 30-degree bank. It follows then that an unopposed yaw damper hard-over would
rapidly increase the bank angle to critical conditions. The first reaction of
- 27 -
the crew would be to decrease the baruc by gradually applying opposxng control
wheel force, probably with a greater delay in reaching full aileron deflection
than the five seconds experienced during previously mentioned test flight
conditions. The pilot may have applied opposite rudder also but with insufficient
force to overpower the servo, with little or no benefit.
The flight recorder traces indicate that five to six seconds after the
malfunction started, the nose -left slip effect of the malfunction suddenly became
greater than the effects of opposing control forces. It can be assumed that
the pilot then applied asymmetric power to arrest the roll, producing the indicated
drop m altitude and the levelling of the airspeed trace at 1008:21 as a result
of decreased sideslip. This power reduction also agrees with the energy analysis.
In conjunction with these altitude and airspeed trace characteristics, consider-
ation of the heading trace indicates the possibility of a time mismatch between
traces, placing the cessation of heading change about one second early. Through-
out this portion of the maneuver the nose -high pitch attitude of the airplane was
maintained. Because of late and inadequate application of lateral control the
momentarily arrested yaw then resumed and started an increasing nose left slip at
time 1008:22, as .indicated by the rising altitude trace.
At 1008:25 the medjan acceleration trace indicates the start of a rapid
increase m load factor to 1.8 g r s at 1008:30. During this rise the individual
deflections of the acceleration trace become higher in frequency than before*
indicating the start of stall buffet. The turbulent airflow over the wing
during stall buffet further decreases the lateral control capability remaining
after lock-out of the outboard ailerons.
It is possible that the increasing load factor progressing to stall buffet
could have been brought about by a combination of some or all of the
following factors: 1. the basic malfunction of the rudder control system was
initially disguised by turbulence and was not quickly identified; 2. the dif-
ficulty of recognizing, in the initial stages, the abnormal attitude of the
aircraft due to excellent YFR conditions tending to decrease frequency of
reference to the attitude instruments; 3- an attempt to maintain the specified
flight departure path as evidenced by the 2.3 nose high elevator trim found m
the wreckage; A- inability to effect immediate corrective action due to
possible initial reliance on lateral control without application of the ad-
ditional effect of speed brakes or flap extension; 5. an unintentional
no sp -high attitude while attempting lateral recovery; 6. the absence of
stick shaker stall warning prior to initial stall buffet; 7 the continued
operation of a malfunctioning yaw damper.
The flight recorder traces suggest that at about 1008:33 the yaw damper
was disengaged, accounting for the sharp decrease in indicated airspeed char-
acteristic of a nose right slip. This leaves sufficient time for retarding
the Nos. 1 and 2 throttles, with resultant reduction of the rpm to flight
idle prior to impact. It appears likely that the rudder boost was deactivated
shortly prior to impact, accounting for the 9 degree right rudder indication
found in examination of the wreckage.
After tune 1008:30 the airplane was m heavy stall buffet, highly abnormal
attitudes, and at altitudes too low for recovery to be effected before crash
- 28 -
The Board therefore concludes that a rudder servo malfunction due to
shorted wires is the most likely abnormality to have produced the accident.
The Board determines that the probable cause of this accident was a
rudder control system malfunction producing yaw, sideslip and roll, leading to
a loss of control from which recovery action was not effective.
The Board presently has made three recommendations to the Administrator
of the Federal Aviation Agency as a result of this accident. The first of
these was that an Airworthiness Directive be issued to require a one-time
inspection of the servo rate generator motors on all Eclipse-Pioneer Model
PB-20D Automatic Flight Control Systems for damaged wire bundles, and that
the Agency take measures as necessary to insure satisfactory quality control
during manufacture and overhaul. The second was that an Airworthiness
Directive be issued to require mandatory incorporation of applicable Boeing
Service Bulletins pertaining to replacement of the Gladden solenoid -operated
valves in the flight control and hydraulic interconnect systems due to flaking
of the nickel plating tending to contaminate the hydraulic fluid. The last
was that the current airworthiness requirements for automatic flight control
systems in Section 4-b. 612(d) of the Civil Air Regulations and the related
CAM material, as specifically applied to the high speed swept-wing design
turbojet aircraft, be reevaluated for the purpose of establishing realistic
time allowances for recognition of abnormal airplane motions, decision to take
corrective action, and initiation of the proper correction in all pertinent
flight regimes; and that necessary changes to the requirements be applied
retroactively to turbojet aircraft equipped with automatic flight control
systems. As of the date of this report the Federal Aviation Agency has taken
appropriate action on the first two recommendations and has the third under
BY THE CIVIL AERONAUTICS BOARD:
/s/ ALAN S. BOYD
/s/ ROBERT T. MURPHY
/s/ CHAN GURNEY
/&/ G. JOSEPH MINETTI
h/ WHITNEY GILLILDAND
The Civil Aeronautics Board was notified of this accident at approximately
1010 on March 1, 1962. Investigators were dispatched immediately to the scene
to conduct an investigation in accordance with the provisions of Title 701(a)(2)
of the Federal Aviation Act of 1958 • A public hearing was ordered by the Board
and held at the International Hotel, New York International Airport, Jamaica,
New "Fork, on March 20-23, 1962 . The investigation was continued until December 1962.
American Airlines, Inc., a Delaware Corporation with General Offices at 633
Third Avenue, New Yorjt City, New York, operates as an air carrier under currently
effective certificates of public convenience and necessity, and an air carrier
operating certificate, both issued pursuant to Federal Aviation Act of 1958, as
amended. These certificates autnorize the transportation by air of persons,
property, and mail between various points in the United States, Mexico, and Canada,
including points on Route it. This route includes, among other cities: New York City,
New York, and Los Angeles, California.
Captain James T. Heist, age 56, was employed by American Airlines, Inc., on
May 1, 19k0 9 and had accumulated a total of 18,300 hours flight time, of which
1,600 hours were in the Boeing 707. He held a currently effective FAA multiengme
land airline transport certificate No, 20152 with numerous ratings, among which
was the Boeing 707 rating. Captain Heist was issued an FAA rating m the Boeing
707 on April 1, I960, and was line qualified on April 25, I960. He received his last
proficiency che'ck in the Boeing 707-123B on October 13, 1961, and his last line
check on September 20, 196l. Records indicate that Captain Heist satisfactorily
passed an FAA first-class flight physical on October 1, 1961, without waivers.
First Officer Michael Barna, Jr., age ?>$> was employed by American Airlines
on January 12, 1953, and had accumulated a total of 1+, 800 hours flight time, of
which 900 hours were in the Boeing 707- He possessed a valid FAA multiengme land
AIR certificate No« 273798 with Douglas DC-6 and DC-7 ratings. Mr. Barna qualified
as first officer on the Boeing 707 on September 30, 1959 • He received his last
proficiency check in the Boeing 720B on December 19, 1961, and his last line check
on February 22, 1962 s in piston equipment. Records indicate that First Officer
Barna satisfactorily passed an FAA first-class flight physical on December 5> 1961,
Second Officer Robert J. Pecor, age 32, was employed by American Airlines on
April 23, 1957, and had accumulated a total of 3,1*00 hours flight time, of which
1,716 were m the Boeing 707* He possessed a valid FAA multiengme land ATR certif-
icate No. 1255374. Mr. Pecor received his last proficiency flight check on May 5*
196l, m a DC-6, and his last line check on August 27, 1961, m a Boe>ng 707-123B.
Records indicate that Mr. Pecor satisfactorily passed an FAA first-class flight
physical on April lk s 1961, without waivers.
Flight Engineer Robert J. Cam, age 32, was employed by American Airlines
on June 16, 195 2 , and had accumulated a total of 7,500 hours flight time, of
which 2,000 hours were in the Boeing 707. He held a valid FAA flight engineer
certificate No. 1245069. Mr. Cain received his last proficiency flight check on
November 28, 1961, and his last line check on December 15, 1961. Records indicate
that Mr. Cain satisfactorily passed an FAA second-class flight physical on Jane 30,
1961, without waivers.
Stewardess Shirley Grabow, age 28, was employed by American Airlines on
December 7, I960.
Stewardess Lois Kelly, age 23, was employed by American Airlines on February 24,
Stewardess Betty Moore, age 22, was employed by American Airlines on
November 17, 1959.
Stewardess Rosalind Stewart, age 20, was employed by American Airlines on
September 12, 1961.
A Boeing 707-123B aircraft, manufacturer's serial No. 17633, U. S. Registry
N 7506A, bore a manufacturer's date of February 12, 1959, and was delivered to
American Airlines, Inc., on the same date.
The last periodic inspection (No. 31) was performed January 18, 1962, when
the TST (Total Ship Time) was 7,922 hours. The second Main Base Check was
accomplished March 26, 1961, with a TST of 5,530 hours. Retrofit, which consisted
of installing Pratt & Whitney JT3D series (Fan) engines, was completed March 3,
1961, and Fin Modification was completed on February 8, 1962. TST as of March 1.
1962, was 8,147 hours. '
The aircraft was powered with four Pratt & Whitney JT3D1 engines with time
since overhaul and total times as follows:
vx b ". l '\ , "! ,; 7'i.'r*' < si
•■> * A «ia\V » f .rM- ' l K»»ooft^ FLAPS2C
h * 527 FT tr^-r^*'"
flaps 20- ' .Snot^
NEW YORK WTKRNATIONAJ-
t - azssci. ■
*? ■ !5 l«
.**• ' IMPACT
27 >*. ->->
« w (3 *«
- 1 JVrD - *0
•^e^rt-, '** „ ~t-/
h-l5fc5FT>* , T5 ,
FLAPS ZCT 0\- '<£
' BUek Wall
3 -.**' H *
.1 « C tt
• •-'■.» ,,1- .
1 ?. 1
y.°* * 3 A^2
Ruffle ««r : ,
tf*f f f
, . . "j&f f * 4m \ 3 \ "js* • • 1^ " wry
*ir a k \- !r;q^€ k a/w a ,v-
'^T » "-«
5 !'.t ^i .i !! ^ ill U
TYPICAL DEPARTURL FLIGHT PMH
AMERICAN AIRLINES 7C7-12 3B
TAKLOFF WE 1081^=247000 LBS- WIND dOKTS-
k- ! : Hf «f h;
CIVIL AERONAUTICS BOARD
FLIGHT RECORDER DATA
AAL BOEING 707-I23B N7506A, JAMAfCA BAY, N.Y, MARCH 1,1962
LAS RECORDER TYPE I09C, SERIAL NUMBER 474
f 006: 50
EASTERN STANDARD TIME
J FP- 12/4/62